相关分析法在NIR检测桑蚕丝含量中的应用

研究采用近红外(NIR)光谱技术快速检测纺织原料中桑蚕丝含量的方法,在用偏最小二乘法(PLS)建立校正模型过程中,探讨互相关分析法对提高建模精度的作用。结果表明:混合数据经过相关分析法处理后,模型的预测精度有所提高,模型的平均绝对误差<2.5(标准差<1.5),测量值与含量参考值具有良好的相关性(相关系数0.996)。近红外光谱快速检测法可以满足桑蚕丝含量的实际测量要求,从而为纺织品的无损、快速检测提供一种新的方法。

Application of correlation analysis to determination of the mulberry silk content by NIR spectroscopy

The method to determine the mulberry silk content in textile by near-infrared (NIR) spectroscopy was studied.The effect of cross-correlation analysis in the quantitative analysis based on partial least squares(PLS) to improve the precision of the model had been proved.The results of blended data show that the predictive precision was enhanced by using cross-correlation analysis,the predicted average absolute error is less than 2.5(the predicted standard error is less than 1.5)and predicted coefficient is more than 0.996.It demonstrates that it is feasible to quickly analyze the mulberry silk content in textile on-line by near-infrared spectroscopy, thus providing a new,nondestructive and fast method for determination of mulberry silk content in textile.